Select the statement that describes the team's current design for each chapter. The results suggest where to start reading. Answers stay in this browser.
1. One round at a glance Can the team trace an extraction problem to a proposed rule change?Not yet The link between an extraction finding and a policy change is unclear. Foundational The team can trace the prepared example from the extraction error to a candidate rule. AI-ready The team records the policy versions, evidence, and decisions behind a revision. AI-native The proposed orchestration design makes those handoffs and checkpoints explicit. 2. Round preparation Can the team reconstruct the conditions of a round?Not yet Data snapshots, references, and versions are scattered. Foundational The starting policy and reviewed examples are documented. AI-ready A manifest pins the batch, schema, settings, and versions. AI-native The runner design checks identities and versions before any model work. 3. Round configuration and startup How would the team start the round and know when to pause?Not yet There is no sequence from the prepared batch to judge findings. Foundational The team can trace the supplied records through extraction and diagnosis. AI-ready The plan specifies which records must be present and which findings pause the round. AI-native The proposed Pi design specifies scoped sessions and recovery from validated records. 4. Human review of findings Can the team resolve flagged findings and check selected unflagged cases?Not yet The team has no human review procedure. Foundational A reviewer compares flagged fields with the image and records decisions. AI-ready The team links escalations, decisions, and spot checks to the affected samples. AI-native The workflow design checks that blocking decisions are recorded and council inputs are scoped. 5. Council deliberation and the next policy How does a finding become a candidate change?Not yet The prompt changes without a decision record. Foundational The team records proposals and the reasons for accepting or rejecting them. AI-ready Independent proposals, critiques, and revisions trace back to evidence and rule IDs. AI-native The workflow design isolates initial proposals and validates the policy and state before handoff. 6. Decisions after a round What evidence would justify stopping development or adopting a policy?Not yet The team relies on informal impressions. Foundational The team tracks rule changes and unresolved issues. AI-ready Stopping criteria and a separate paired quality comparison are specified. AI-native The design links versioned evidence to independent evaluation and an adoption decision. 0 of 6 answered Show reading path Reset